Reasoning Fidelity on AIME 2024 (ref6)
91.7Top-1 AccCASK
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CASKBudget=5122026.04 | 91.7 | 99.9 | 0.233 | 8.7 | 43.5 | |
| CASKBudget=3842026.04 | 90.7 | 99.7 | 0.268 | 8.7 | 61.6 | |
| TriAttentionBudget=5122026.04 | 89.4 | 99.1 | 0.333 | 8.7 | 43.6 | |
| CASKBudget=2562026.04 | 88.4 | 99.2 | 0.359 | 8.7 | 65.3 | |
| TriAttentionBudget=3842026.04 | 88.2 | 98.9 | 0.383 | 8.7 | 61.6 | |
| TriAttentionBudget=2562026.04 | 86.1 | 98 | 0.463 | 8.7 | 65.3 |